The New York Times has an interesting piece on this here. I’m all in favor of this type of approach as long as it’s rigorously quantified. The key idea is the it’s driven by photosynthesis and is essentially free. On the other hand, the question becomes how does one optimize the phenotype to achieve this goal?